Electrical operation misoperation prediction and correction system based on artificial intelligence

Abstract

The invention relates to the technical field of electrical operation safety management, in particular to an artificial intelligence-based electrical operation misoperation prediction and correction system, which comprises a data acquisition module, a data processing module, a misoperation prediction module, a correction suggestion generation module, a real-time monitoring and feedback module, a self-learning module and a data storage module, operation data and environment data are collected in real time through a sensor network, the data are preprocessed, feature vectors are generated, a misoperation prediction model is constructed based on a long short-term memory network algorithm, the misoperation occurrence probability is predicted, a decision tree algorithm is adopted to generate correction suggestions, and prediction results and the correction suggestions are presented through a human-computer interaction interface. The self-learning module is used for dynamically optimizing the prediction model and the suggestion generation model; the data storage module is used for classifying and storing the operation data, the environment data and the prediction result; according to the invention, the misoperation is accurately predicted in real time, the intelligent correction suggestion is generated, and the safety and efficiency of electrical operation are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical operation safety management, and in particular to an electrical operation misoperation prediction and correction system based on artificial intelligence. Background Art

[0002] With the acceleration of industrialization and the widespread application of electrical equipment, electrical work plays a vital role in industrial production and daily life. However, due to the complex environment, frequent operations and high safety requirements of electrical work, the slightest carelessness during the operation may cause serious accidents such as equipment failure or personal injury. Therefore, the safety of electrical work has become an important issue in industrial production.

[0003] Currently, many electrical operation management systems rely on data collection and analysis technology, combined with artificial intelligence algorithms for operation optimization. However, the existing systems have obvious deficiencies in the real-time performance of data collection, the accuracy of prediction models, and the intelligent level of suggestion generation, and cannot fully meet the high safety and efficiency requirements in complex operation scenarios. For example, data acquisition equipment can usually only cover limited operating parameters or environmental data, resulting in insufficient data dimensions and poor real-time performance; the prediction model has limited performance in dealing with multi-source data fusion and dynamic feature extraction, and it is difficult to achieve accurate prediction of misoperation. In addition, the suggestion generation module lacks intelligent sorting and optimization mechanisms, and cannot provide operators with clear and actionable decision support. Summary of the invention

[0004] Based on the above objectives, the present invention provides an electrical operation error operation prediction and correction system based on artificial intelligence.

[0005] An artificial intelligence-based electrical operation misoperation prediction and correction system includes a data acquisition module, a data processing module, a misoperation prediction module, a correction suggestion generation module, a real-time monitoring and feedback module, a self-learning module and a data storage module, wherein;

[0006] The data acquisition module collects operation data and environmental data in the electrical operation process in real time through the sensor network. The operation data includes the operation parameters of the operator (such as key actions, equipment switch status, operation frequency, etc.) and the operating status of the electrical equipment (such as equipment load, voltage, current, power factor, etc.). The environmental data includes ambient temperature, humidity, noise level, light intensity and air quality.

[0007] The data processing module is used to pre-process the collected operation data and environmental data, including data cleaning, outlier detection and feature extraction;

[0008] The misoperation prediction module constructs a misoperation prediction model based on a long short-term memory network algorithm, and uses historical data and real-time data to predict the probability of misoperation. The historical data includes historical operation data and historical environment data, and the real-time data includes real-time operation data and real-time environment data.

[0009] The correction suggestion generation module constructs a correction suggestion generation model through a decision tree algorithm according to the output of the misoperation prediction module, and generates corresponding correction suggestions through the correction suggestion generation model. The correction suggestions include operation step adjustment, equipment parameter optimization and safety warning;

[0010] The real-time monitoring and feedback module is used to monitor the electrical operation process in real time and feed back the prediction results and correction suggestions to the operator. The real-time monitoring and feedback module includes a human-computer interaction interface, which is used to present the misoperation prediction information and correction suggestions to the operator in a visual form and provide a real-time alarm function;

[0011] The self-learning module dynamically optimizes the error operation prediction model and the correction suggestion generation model through a reinforcement learning algorithm. The self-learning module performs reinforcement learning based on a Q-learning algorithm and dynamically updates the error operation prediction model and the correction suggestion generation model according to the operator's response to the correction suggestion.

[0012] The data storage module is used to store the operation data, environmental data, error operation prediction results and correction suggestions during the operation process in the cloud database.

[0013] Optionally, the data acquisition module specifically includes:

[0014] Sensor deployment: Multiple types of sensors are deployed at the electrical operation site to collect operation data and environmental data. The sensors include pressure sensors, displacement sensors, Hall effect sensors, power sensors, current and voltage sensors, temperature and humidity sensors, noise sensors, light sensors, and PM2.5 sensors.

[0015] Operation data collection: The operator's operating parameters, the pressure sensor and displacement sensor are used to record the pressing strength and displacement of the key action to analyze the operating habits and identify abnormal actions. The Hall effect sensor is used to detect the switch state of the equipment, and the state value is marked as 0 or 1. The counter records the number of operations per unit time in real time (unit: times / minute);

[0016] The operating status of electrical equipment uses power sensors to collect equipment load values, current and voltage sensors to monitor the voltage and current during operation, and power analyzers to record power factors (ranging from 0 to 1);

[0017] Environmental data collection: The temperature and humidity of the working environment are monitored in real time through temperature and humidity sensors, the noise level of the working environment is recorded through noise sensors, the light intensity of the work site is collected through photosensors, and the air quality is monitored in real time through PM2.5 sensors, so as to accurately grasp the environmental conditions of the work site.

[0018] Optionally, the data processing module includes:

[0019] Data cleaning: Clean the operation data and environment data. Calibrate the operation data and environment data through the time synchronization mechanism to ensure that the data records have a unified timestamp. Use linear interpolation to fill in missing values ​​in the operation data and environment data to ensure data integrity. Use rule filtering to remove duplicate data or data in abnormal formats to ensure that the cleaned data conforms to the expected format and has logical consistency.

[0020] Outlier detection: Use the IQR (interquartile range) method to detect outliers in operational data and environmental data;

[0021] Feature extraction: Extract features from the cleaned data based on the multimodal data fusion algorithm to generate a unified feature vector.

[0022] Optionally, the misoperation prediction module specifically includes:

[0023] Data preparation: Preprocess historical data and real-time data, and divide the feature vectors generated by the data processing module into training sets and test sets. The training set is used for model training, and the test set is used for model verification.

[0024] Model construction: A misoperation prediction model is constructed based on the long short-term memory network algorithm. The core of the model includes an input layer, multiple long short-term memory network hidden layers, and an output layer.

[0025] Model training: Use the training set data to train the constructed error operation prediction model, and use the cross entropy loss function to measure the prediction error of the model;

[0026] Real-time prediction: After the model training is completed, the real-time data feature vector is input into the trained error operation prediction model to predict the error operation probability P(e t ), if the predicted value exceeds the preset threshold, a warning message is sent to the correction suggestion generation module.

[0027] Optionally, the correction suggestion generating module includes:

[0028] Data input: Receive the predicted probability of erroneous operation P(e) from the erroneous operation prediction module t) and related characteristic data as input, the characteristic data includes operation data (such as key actions, device switch status, operation frequency, etc.) and environmental data (such as temperature, humidity, noise level, etc.);

[0029] Decision tree construction: A correction suggestion generation model is constructed through a decision tree algorithm, and the correction suggestion generation model is trained based on input data and its corresponding labels (normal operation or incorrect operation);

[0030] Model optimization: The decision tree is pruned through cross-validation method to remove redundant nodes.

[0031] Optionally, the correction suggestion generating module further includes:

[0032] Suggestion generation: Generate specific corrective suggestions based on the output categories of the corrective suggestion generation model;

[0033] Suggestion prioritization: Prioritize the generated correction suggestions;

[0034] Suggestion output: The generated corrective suggestions are passed to the real-time monitoring and feedback module and displayed in a visual form through the human-computer interaction interface.

[0035] Optionally, the real-time monitoring and feedback module includes:

[0036] Data reception: Receive the prediction results of the occurrence of error operations from the error operation prediction module, including the prediction probability P(e t ), receiving the generated correction suggestion from the correction suggestion generating module;

[0037] Data analysis and processing: Analyze the received prediction results and correction suggestions, and convert the prediction probability P(e t ) is converted into easily understandable risk levels, which are divided into high risk, medium risk and low risk;

[0038] Visual presentation: Using the human-computer interaction interface, real-time monitoring data and corrective suggestions are presented to operators in a visual form;

[0039] Real-time alarm: When the prediction results show high risk (P(e t )>0.8), the alarm function is triggered;

[0040] User interaction: Operators provide feedback on corrective suggestions through the human-computer interaction interface.

[0041] Optionally, the self-learning module:

[0042] Environment modeling: The self-learning module builds a reinforcement learning environment based on real-time data and user feedback, defining states, actions, and rewards;

[0043] Q-learning algorithm initialization: Define the value function Q(s,a) of Q-learning, which represents the long-term reward that can be obtained by selecting action a in state s, expressed as:

[0044]

[0045] Among them, α is the learning rate, γ is the discount factor, r is the immediate reward, and s ′ is the new state after action a is executed, is the maximum Q value of the next state;

[0046] Model optimization: Dynamically adjust the error operation prediction model and correction suggestion generation model based on reinforcement learning strategies;

[0047] Strategy iteration: The self-learning module adopts ∈-greedy strategy for action selection;

[0048] User feedback loop: Collect the actual implementation of corrective suggestions by operators, including feedback results (whether the suggestions are adopted and the actual effects) and changes in system status after the operation, continuously update the training environment of reinforcement learning, and form a closed-loop optimization.

[0049] Optionally, the data storage module includes:

[0050] Data classification and labeling: Classify and store the operation data, environmental data, error operation prediction results and correction suggestions provided by the data acquisition module, error operation prediction module and correction suggestion generation module;

[0051] Data formatting: Format the data before storing it;

[0052] Data compression and encryption: compress and encrypt data to improve storage efficiency and security;

[0053] Data transmission and storage: The compressed and encrypted data is uploaded to the cloud database for storage via a secure transmission protocol (such as HTTPS or MQTT protocol).

[0054] Beneficial effects of the present invention:

[0055] The present invention, through the collaborative work of the data acquisition module and the data processing module, can collect the operation data and environmental data in the operation process in real time, and clean the data, detect outliers and extract features to ensure the accuracy and integrity of the data. Based on this, the efficiently constructed misoperation prediction model uses the long short-term memory network algorithm to accurately capture the dynamic characteristics of time series data, and combines real-time data and historical data to achieve accurate prediction of the probability of misoperation. The combination of real-time and accuracy significantly reduces the incidence of misoperation and provides a solid guarantee for operation safety.

[0056] In the present invention, the corrective suggestion generation module constructs a model through a decision tree algorithm, and can generate specific operation step adjustments, equipment parameter optimization and safety warning suggestions based on the prediction results, and prioritize the suggestions to ensure that the operators receive effective and clear guidance in a timely manner. Combined with the data visualization and real-time alarm functions of the real-time monitoring and feedback module, the operators can quickly understand the current operation risks and adopt targeted corrective suggestions to effectively prevent potential safety hazards. In addition, the self-learning module dynamically optimizes the prediction model and the suggestion generation model through a reinforcement learning algorithm, realizes the adaptive improvement of the system according to the feedback of the operators, and significantly improves the intelligent level of suggestion generation.

[0057] In the present invention, the data storage module classifies, formats, encrypts, compresses and stores hierarchically the operation data, environmental data, error operation prediction results and correction suggestions collected during the operation, ensuring the security, integrity and easy retrieval of the data. Through the long-term storage and life cycle management mechanism of the cloud database, historical data can not only be used for trend analysis and model optimization, but also provide key support for safety audits and accident analysis, further improving the long-term effectiveness and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 A schematic diagram of a system flow of an embodiment of the present invention;

[0060] Figure 2 Schematic diagram of the flow of the misoperation prediction module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0062] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0063] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0064] like Figure 1-Figure 2 As shown, an artificial intelligence-based electrical operation misoperation prediction and correction system includes a data acquisition module, a data processing module, a misoperation prediction module, a correction suggestion generation module, a real-time monitoring and feedback module, a self-learning module and a data storage module, wherein;

[0065] The data acquisition module collects the operation data and environmental data in the electrical operation process in real time through the sensor network. The operation data includes the operation parameters of the operator (such as key actions, equipment switch status, operation frequency, etc.) and the operation status of the electrical equipment (such as equipment load, voltage, current, power factor, etc.). The environmental data includes ambient temperature, humidity, noise level, light intensity and air quality.

[0066] The data processing module is used to pre-process the collected operation data and environmental data, including data cleaning, outlier detection and feature extraction;

[0067] The error operation prediction module builds an error operation prediction model based on the long short-term memory network algorithm, and uses historical data and real-time data to predict the probability of error operation. The historical data includes historical operation data and historical environment data, and the real-time data includes real-time operation data and real-time environment data.

[0068] The correction suggestion generation module builds a correction suggestion generation model through a decision tree algorithm based on the output of the error operation prediction module, and generates corresponding correction suggestions through the correction suggestion generation model. The correction suggestions include operation step adjustment, equipment parameter optimization and safety warning;

[0069] The real-time monitoring and feedback module is used to monitor the electrical operation process in real time and provide feedback of the prediction results and correction suggestions to the operators. The real-time monitoring and feedback module includes a human-computer interaction interface, which is used to present the error operation prediction information and correction suggestions to the operators in a visual form and provide a real-time alarm function;

[0070] The self-learning module dynamically optimizes the error operation prediction model and the correction suggestion generation model through the reinforcement learning algorithm. The self-learning module performs reinforcement learning based on the Q-learning algorithm and dynamically updates the error operation prediction model and the correction suggestion generation model according to the operator's response to the correction suggestion.

[0071] The data storage module is used to store the operation data, environmental data, error operation prediction results and correction suggestions during the operation process in the cloud database.

[0072] The data acquisition module specifically includes:

[0073] Sensor deployment: Multiple types of sensors are deployed at the electrical operation site to collect operation data and environmental data. The sensors include pressure sensors, displacement sensors, Hall effect sensors, power sensors, current and voltage sensors, temperature and humidity sensors, noise sensors, light sensors, and PM2.5 sensors.

[0074] Operation data collection: The operator's operating parameters, the pressure sensor and displacement sensor are used to record the pressing strength and displacement of the key action to analyze the operating habits and identify abnormal actions. The Hall effect sensor is used to detect the switch state of the equipment, and the state value is marked as 0 or 1. The counter records the number of operations per unit time in real time (unit: times / minute);

[0075] The operating status of electrical equipment uses power sensors to collect equipment load values, current and voltage sensors to monitor the voltage and current during operation, and power analyzers to record power factors (ranging from 0 to 1);

[0076] Environmental data collection: The temperature and humidity of the working environment are monitored in real time through temperature and humidity sensors, the noise level of the working environment is recorded through noise sensors, the light intensity of the work site is collected through photosensors, and the air quality is monitored in real time through PM2.5 sensors, so as to accurately grasp the environmental conditions of the work site.

[0077] The data processing module includes:

[0078] Data cleaning: Clean the operation data and environment data. Calibrate the operation data and environment data through the time synchronization mechanism to ensure that the data records have a unified timestamp. Use linear interpolation to fill in missing values ​​in the operation data and environment data to ensure data integrity. Use rule filtering to remove duplicate data or data in abnormal formats to ensure that the cleaned data conforms to the expected format and has logical consistency.

[0079] Outlier detection: Use the IQR (interquartile range) method to detect outliers in operational data and environmental data. The specific steps include:

[0080] (1) Calculate the first quartile (Q1) and the third quartile (Q3) of the data, and obtain the interquartile range (IQR), the formula is:

[0081] IQR = Q3-Q1;

[0082] (2) The normal range of the data was determined to be [Q1-1.5×IQR, Q3+1.5×IQR], and values ​​outside this range were marked as outliers;

[0083] (3) For detected outliers, they are processed by referring to the data distribution trend. If the error is small, the median is used as a replacement; if the error is large, it is directly eliminated;

[0084] Feature extraction: Extract features from the cleaned data based on the multimodal data fusion algorithm to generate a unified feature vector. The specific steps are as follows:

[0085] (1) Normalization: Normalize the data of different physical quantities using the Min-Max normalization formula, expressed as:

[0086]

[0087] Among them, x is the original data, min(X) and max(X) are the minimum and maximum values ​​of the data set respectively, and x ′ is the normalized data;

[0088] (2) Feature combination: The multi-source data is fused into a unified feature vector through the weighted average fusion algorithm, which is expressed as:

[0089] X=w1aX1+w2·X2+…+w n ·X n ;

[0090] Among them, X i is the feature vector of the ith data source, w i is the corresponding weight, satisfying Weight w iAssign values ​​based on the importance of the data through expert experience or statistical analysis;

[0091] The data processing module efficiently converts raw data into standardized and structured feature vectors through three steps: data cleaning, outlier detection and feature extraction, providing high-quality input data for the subsequent error operation prediction model. This processing method can significantly improve the prediction accuracy of the model while enhancing the robustness of the system in complex data environments.

[0092] The error operation prediction module specifically includes:

[0093] Data preparation: Preprocess historical data and real-time data, and divide the feature vectors generated by the data processing module into training sets and test sets. The training set is used for model training, and the test set is used for model verification. The historical data includes the historical operating parameters of operators, the historical operating status of electrical equipment, the historical environmental parameters of the operating environment, and the misoperation mark data;

[0094] Model construction: Based on the long short-term memory network algorithm, an error operation prediction model is constructed. The core of the model includes an input layer, multiple long short-term memory network hidden layers and an output layer, among which;

[0095] (1) Input layer: accepts standardized data feature vectors in the form of time series data, expressed as:

[0096] X=[x1,x2,…,x T ];

[0097] Among them, x t The feature vector representing time step t;

[0098] (2) Long short-term memory network hidden layer: used to capture the short-term and long-term dependencies of time series data. The hidden state update formula is as follows:

[0099] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0100] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0101]

[0102] h t =o t tanh(C t );

[0103] Among them, f t 、i t and t They are forget gate, input gate and output gate respectively, C t is the cell state, h t is the hidden state, σ represents the Sigmoid activation function, tanh represents the hyperbolic tangent activation function, W f , W i , W C is the weight matrix, b f 、b u 、b C is the bias vector;

[0104] (3) Output layer: Output the probability of an erroneous operation P(e) through the Softmax function t ), expressed as:

[0105]

[0106] Among them, y t is the output value of the tth time step, K is the total number of classification categories;

[0107] Model training: Use the training set data to train the constructed error operation prediction model, and use the cross entropy loss function L to measure the prediction error of the model, which is defined as follows:

[0108]

[0109] Among them, y i is the actual label, is the predicted probability, N is the total number of samples, and the model parameters are iteratively updated through the back-propagation algorithm and optimizer (such as Adam) until the loss function converges;

[0110] Real-time prediction: After the model training is completed, the real-time data feature vector is input into the trained error operation prediction model to predict the error operation probability P(e t ), if the predicted value exceeds the preset threshold, a warning message is sent to the correction suggestion generation module;

[0111] The error operation prediction module captures the dynamic characteristics in time series data through the long short-term memory network algorithm, combines the long-term information of historical data and the latest changes of real-time data, and realizes accurate prediction of the probability of error operation, which significantly improves the prediction accuracy of the system and provides a reliable basis for the corrective decision-making of subsequent modules.

[0112] The correction suggestion generation module includes:

[0113] Data input: Receive the predicted probability of erroneous operation P(e) from the erroneous operation prediction module t ) and related characteristic data as input, the characteristic data includes operation data (such as key actions, device switch status, operation frequency, etc.) and environmental data (such as temperature, humidity, noise level, etc.);

[0114] Decision tree construction: A correction suggestion generation model is constructed through a decision tree algorithm. The correction suggestion generation model is trained based on input data and its corresponding labels (normal operation or incorrect operation). The specific steps include:

[0115] Feature selection: Use information gain IG to determine the importance of each feature. The formula is:

[0116]

[0117] Among them, H(T) is entropy, A is feature, T v represents a subset of feature values ​​v;

[0118] Split node: Select the feature with the largest information gain as the split node to build a tree structure;

[0119] Recursive splitting: Repeat the above process until the preset tree depth is reached or the information gain is less than the set threshold;

[0120] Leaf node generation: Output corrective suggestion categories at leaf nodes, including “operation step adjustment”, “equipment parameter optimization” and “safety warning”;

[0121] Model optimization: Prune the decision tree through cross-validation method to remove redundant nodes, prevent overfitting, and improve the generalization ability of the model;

[0122] The corrective suggestion generation module can efficiently generate corrective suggestion categories suitable for different operating scenarios by building and optimizing the corrective suggestion generation model, providing operators with targeted safety decision support.

[0123] The correction suggestion generation module also includes:

[0124] Suggestion generation: Generates specific corrective suggestions based on the output categories of the corrective suggestion generation model, which are divided into the following three types:

[0125] (1) Operation step adjustment: When the decision tree outputs the "operation step adjustment" category, a specific operation adjustment plan is generated in combination with the current operation data, for example:

[0126] Provide optimization suggestions for the operation sequence (e.g. shut down device A first, then operate device B);

[0127] Prompt operators to reduce operation frequency or optimize operation intensity;

[0128] (2) Equipment parameter optimization: When the decision tree outputs the "equipment parameter optimization" category, it provides specific equipment parameter adjustment suggestions, such as:

[0129] Adjust the equipment operating load to a safe range (e.g. load from 80% to 60%);

[0130] Optimize the voltage or current value of the equipment;

[0131] (3) Safety warning: When the decision tree outputs the “safety warning” category, a warning message is generated, for example:

[0132] Prompts that the operation is high-risk (for example, pressing the key too quickly may cause accidental touches);

[0133] The environment is not suitable for work (e.g. excessive noise may affect concentration);

[0134] Suggestion Prioritization: Prioritize the generated correction suggestions according to the following rules:

[0135] High priority: Security warning suggestions have the highest priority;

[0136] Medium priority: Device parameter optimization suggestions have medium priority;

[0137] Low priority: Suggestions for adjusting operation steps have the lowest priority;

[0138] Suggestion output: The generated corrective suggestions are transmitted to the real-time monitoring and feedback module and displayed in a visual form through the human-computer interaction interface to ensure that operators can quickly understand and execute them;

[0139] The corrective suggestion generation module ensures that the generated suggestions are highly targeted and actionable by refining the suggestion content and prioritizing them, providing comprehensive and practical support to operators, thereby significantly reducing the risk of incorrect operations.

[0140] Real-time monitoring and feedback modules include:

[0141] Data reception: Receive the prediction results of the occurrence of error operations from the error operation prediction module, including the prediction probability P(e t ), receiving the generated correction suggestion from the correction suggestion generating module;

[0142] Data analysis and processing: Analyze the received prediction results and correction suggestions, and convert the prediction probability P(e t ) is converted into easily understandable risk levels, which are divided into high risk, medium risk and low risk as follows:

[0143] P(e t )>0.8: high risk (marked as red warning);

[0144] 0.5≤P(e t )≤0.8: medium risk (marked as yellow warning);

[0145] P(e t )<0.5: low risk (marked with green hint);

[0146] Sort the corrective suggestions by priority, ensuring that high-priority suggestions are displayed first;

[0147] Visual presentation: Using the human-computer interaction interface, real-time monitoring data and corrective suggestions are presented to operators in a visual form, including:

[0148] Real-time monitoring data: Displays the predicted probability of misoperation, corresponding risk level, operating parameters (such as key actions, equipment operating status, etc.) and environmental parameters (such as temperature, humidity, noise level, etc.);

[0149] Corrective suggestions: Specific suggestions (such as operating step adjustments, equipment parameter optimization, and safety warnings) are presented in the form of a step list or parameter adjustment table.

[0150] Graphical display: intuitively present the risk change trend through bar charts, heat maps, etc.

[0151] Real-time alarm: When the prediction results show high risk (P(e t )>0.8), the alarm function is triggered, including the following forms:

[0152] Sound and light alarm: remind operators to pay attention to safety through buzzer and red flashing light;

[0153] Screen warning: relevant information is highlighted on the human-computer interaction interface, and a warning box pops up to indicate risks and suggestions;

[0154] User interaction: Operators provide feedback on correction suggestions through the human-computer interaction interface, including:

[0155] Suggestion confirmation: The operator chooses whether to adopt the corrective suggestion and marks the validity of the suggestion;

[0156] Data submission: The feedback results are recorded in the data storage module for use by the self-learning module to optimize the model;

[0157] The real-time monitoring and feedback module ensures that operators can promptly grasp the prediction results and correction suggestions for misoperation through data reception, analysis and processing, visual display and real-time alarm. It also improves the feedback efficiency and intelligence level of the system through user interaction, thereby enhancing the system's real-time response capability and operational safety.

[0158] Self-learning modules:

[0159] Environment modeling: The self-learning module builds a reinforcement learning environment based on real-time data and user feedback, defining states, actions, and rewards, including:

[0160] State (S): including the error operation prediction result P(e t ), real-time operating parameters (such as key actions, device switch status, etc.), environmental parameters (such as temperature, humidity, etc.) and user feedback information;

[0161] Action (A): includes dynamically adjusting the parameters of the error prediction model and optimizing the rules of the correction suggestion generation model, such as updating the error prediction model weights or adjusting the decision tree splitting threshold;

[0162] Reward (R): Generates a reward value based on the operator's feedback on the correction suggestion:

[0163] If the suggestion is adopted and effective, the reward is positive (e.g. R = +1);

[0164] If the suggestion is rejected or invalid, the reward is negative (e.g. R = -1);

[0165] Q-learning algorithm initialization: Define the value function Q(s,a) of Q-learning, which represents the long-term reward that can be obtained by selecting action a in state s, expressed as:

[0166]

[0167] Among them, α is the learning rate, γ is the discount factor, r is the immediate reward, and s ′ is the new state after action a is executed, is the maximum Q value of the next state;

[0168] Model optimization: Dynamically adjust the error operation prediction model and correction suggestion generation model based on reinforcement learning strategies, including:

[0169] Error prediction model optimization: Regularly retrain the long short-term memory network of the error prediction model, expand the training set through collected feedback data, and update the model parameters to improve prediction accuracy;

[0170] Corrective suggestion generation model optimization: Dynamically update the splitting rules or pruning strategies of the corrective suggestion generation model, and adjust the feature selection weights of node splitting according to the reward function to optimize the accuracy and effectiveness of the suggestions;

[0171] Strategy iteration: The self-learning module uses the ∈-greedy strategy for action selection, including:

[0172] Randomly select actions with probability ∈ (exploration) to improve model adaptability;

[0173] Select the action with the largest current Q value (utilization) with probability 1-∈ to improve efficiency;

[0174] As the learning process progresses, the value of ∈ is gradually reduced and the utilization behavior is increased;

[0175] User feedback loop: collects the actual implementation of corrective suggestions by operators, including feedback results (whether the suggestions are adopted and the actual effects) and changes in system status after the operation, continuously updates the training environment of reinforcement learning, and forms a closed-loop optimization;

[0176] The self-learning module continuously improves the accuracy and adaptability of the error operation prediction model and the correction suggestion generation model through the dynamic optimization mechanism of reinforcement learning and Q-learning algorithm. It builds an adaptive closed-loop learning system using the feedback information of operators, which significantly improves the intelligence level of the system and the reliability of long-term operation.

[0177] The data storage module includes:

[0178] Data classification and labeling: The operation data, environmental data, error operation prediction results and correction suggestions provided by the data acquisition module, error operation prediction module and correction suggestion generation module are classified and stored, including:

[0179] Operation data: including the operator's operating parameters (such as keystrokes, equipment switch status, operating frequency, etc.) and electrical equipment operating status (such as equipment load, voltage, current, power factor, etc.);

[0180] Environmental data: including temperature, humidity, noise level, light intensity and air quality at the work site;

[0181] Error operation prediction results: including the predicted probability of error operation and related feature information;

[0182] Corrective suggestions: including specific suggestions and priority labels for operating procedure adjustments, equipment parameter optimization, and safety warnings;

[0183] Data formatting: Data is formatted before storage to ensure that the data structure is unified and easy to store, including:

[0184] Timestamp unification: Add standardized timestamps to each data record to ensure that all data is stored in chronological order;

[0185] Data standardization: convert data from different sources (such as electrical parameters, environmental parameters) into a unified storage format (such as JSON or Parquet format);

[0186] Data integrity check: A check mechanism is used to ensure that there is no loss or damage during data transmission. If abnormal data is found, a retransmission mechanism is triggered;

[0187] Data compression and encryption: compress and encrypt data to improve storage efficiency and security;

[0188] Data transmission and storage: The compressed and encrypted data is uploaded to the cloud database for storage via a secure transmission protocol (such as HTTPS or MQTT protocol). The storage structure includes:

[0189] Tiered storage: Divide storage layers according to data types (operational data, environmental data, prediction results, correction suggestions) to facilitate query and management;

[0190] Index building: Create a unique index for each data record to support fast retrieval.

[0191] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0192] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An artificial intelligence-based electrical operation error prediction and correction system, characterized in that: It includes data acquisition module, data processing module, misoperation prediction module, correction suggestion generation module, real-time monitoring and feedback module, self-learning module and data storage module, among which; The data acquisition module collects operation data and environmental data in the electrical operation process in real time through a sensor network. The operation data includes the operation parameters of the operator and the operating status of the electrical equipment. The environmental data includes the ambient temperature, humidity, noise level, light intensity and air quality. The data processing module is used to pre-process the collected operation data and environmental data, including data cleaning, outlier detection and feature extraction; The error operation prediction module constructs an error operation prediction model based on the long short-term memory network algorithm, and uses historical data and real-time data to predict the probability of error operation. The correction suggestion generation module constructs a correction suggestion generation model through a decision tree algorithm according to the output of the misoperation prediction module, and generates corresponding correction suggestions through the correction suggestion generation model. The correction suggestions include operation step adjustment, equipment parameter optimization and safety warning; The real-time monitoring and feedback module is used to monitor the electrical operation process in real time and feed back the prediction results and correction suggestions to the operator. The real-time monitoring and feedback module includes a human-computer interaction interface, which is used to present the misoperation prediction information and correction suggestions to the operator in a visual form and provide a real-time alarm function; The self-learning module dynamically optimizes the error operation prediction model and the correction suggestion generation model through a reinforcement learning algorithm. The self-learning module performs reinforcement learning based on a Q-learning algorithm and dynamically updates the error operation prediction model and the correction suggestion generation model according to the operator's response to the correction suggestion. The data storage module is used to store the operation data, environmental data, error operation prediction results and correction suggestions during the operation process in the cloud database.

2. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 1, characterized in that: The data acquisition module specifically includes: Sensor deployment: Multiple types of sensors are deployed at the electrical operation site to collect operation data and environmental data. The sensors include pressure sensors, displacement sensors, Hall effect sensors, power sensors, current and voltage sensors, temperature and humidity sensors, noise sensors, light sensors, and PM2.5 sensors. Operation data collection: The operator's operating parameters, the pressure sensor and displacement sensor are used to record the pressing strength and displacement of the key action, the Hall effect sensor is used to detect the switch status of the equipment, and the counter is used to record the number of operations per unit time in real time; The operating status of electrical equipment uses power sensors to collect equipment load values, current and voltage sensors to monitor the voltage and current during operation, and power analyzers to record power factors; Environmental data collection: The temperature and humidity of the working environment are monitored in real time through temperature and humidity sensors, the noise level of the working environment is recorded through noise sensors, the light intensity of the work site is collected through photosensitive sensors, and the air quality is monitored in real time through PM2.5 sensors.

3. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 2 is characterized in that: The data processing module comprises: Data cleaning: Clean the operation data and environment data. Calibrate the operation data and environment data through the time synchronization mechanism. Use linear interpolation to fill in missing values ​​in the operation data and environment data. Use rule filtering to remove duplicate data or data in abnormal formats. Outlier detection: Use the IQR method to detect outliers in operational data and environmental data; Feature extraction: Extract features from the cleaned data based on the multimodal data fusion algorithm to generate a unified feature vector.

4. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 3 is characterized in that: The misoperation prediction module specifically includes: Data preparation: Preprocess historical data and real-time data, and divide the feature vectors generated by the data processing module into training sets and test sets. The training set is used for model training, and the test set is used for model verification. Model construction: A misoperation prediction model is constructed based on the long short-term memory network algorithm. The core of the model includes an input layer, multiple long short-term memory network hidden layers, and an output layer. Model training: Use the training set data to train the constructed error operation prediction model, and use the cross entropy loss function to measure the prediction error of the model; Real-time prediction: After the model training is completed, the real-time data feature vector is input into the trained error operation prediction model to predict the probability of error operation at the current time step. If the predicted value exceeds the preset threshold, an early warning message is sent to the correction suggestion generation module.

5. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 4 is characterized in that: The correction suggestion generating module comprises: Data input: receiving the predicted probability of occurrence of erroneous operation and related feature data from the erroneous operation prediction module as input, the feature data including operation data and environment data; Decision tree construction: Build a correction suggestion generation model through the decision tree algorithm, and train the correction suggestion generation model based on the input data and its corresponding labels; Model optimization: The decision tree is pruned through cross-validation method to remove redundant nodes.

6. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 5, characterized in that: The correction suggestion generating module also includes: Suggestion generation: Generates corrective suggestions based on the output categories of the corrective suggestion generation model; Suggestion prioritization: Prioritize the generated correction suggestions; Suggestion output: The generated corrective suggestions are passed to the real-time monitoring and feedback module and displayed in a visual form through the human-computer interaction interface.

7. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 6, characterized in that: The real-time monitoring and feedback module includes: Data reception: receiving the prediction results of the occurrence of erroneous operations from the erroneous operation prediction module, including the prediction probability, and receiving the generated correction suggestions from the correction suggestion generation module; Data analysis and processing: Analyze the received prediction results and correction suggestions, convert the prediction probability into risk level, and divide the risk level into high risk, medium risk and low risk; Visual presentation: Using the human-computer interaction interface, real-time monitoring data and corrective suggestions are presented to operators in a visual form; Real-time alarm: When the prediction results show high risk, the alarm function is triggered; User interaction: Operators provide feedback on corrective suggestions through the human-computer interaction interface.

8. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 7, characterized in that: The self-learning module: Environment modeling: The self-learning module builds a reinforcement learning environment based on real-time data and user feedback, defining states, actions, and rewards; Q-learning algorithm initialization: Define the value function Q(s,a) of Q-learning, which represents the long-term reward that can be obtained by selecting action a in state s, expressed as: Among them, α is the learning rate, γ is the discount factor, r is the immediate reward, and s ′ is the new state after action a is executed, max a′ Q(s ′ ,a ′ ) is the maximum Q value of the next state; Model optimization: Dynamically adjust the error operation prediction model and correction suggestion generation model based on reinforcement learning strategies; Strategy iteration: The self-learning module adopts ∈-greedy strategy for action selection; User feedback loop: Collect the actual implementation of corrective suggestions by operators, including feedback results and changes in system status after operation, and continuously update the training environment of reinforcement learning to form a closed-loop optimization.

9. The artificial intelligence-based electrical operation misoperation prediction and correction system according to claim 8, characterized in that: The data storage module comprises: Data classification and labeling: Classify and store the operation data, environmental data, error operation prediction results and correction suggestions provided by the data acquisition module, error operation prediction module and correction suggestion generation module; Data formatting: Format the data before storing it; Data compression and encryption: compress and encrypt data to improve storage efficiency and security; Data transmission and storage: The compressed and encrypted data is uploaded to the cloud database for storage via a secure transmission protocol.

Images